S-Wave Velocity Images of the Crust in the Southeast Margin of Tibet Revealed by Receiver Functions

Abstract

The southeast margin of Tibet is the region in clockwise rotation around the Eastern Himalayan Syntaxis due to the India–Eurasia collision and the resistance of the stable Sichuan Basin and South China block. However, the dynamic processes involved in the evolution and deformation of the region remain poorly understood due to a lack of reliable geophysical observations. We collected abundant seismic data recorded by 108 permanent broadband stations deployed in the SE margin of Tibet since 2000, and obtained 4536 pairs of P-wave receiver functions (PRFs) with high signal-to-noise ratio. In this study, we have implemented a novel two-step data inversion procedure that can reduce the dependence of the inversion results on the initial model. We first use low-frequency PRFs obtained by iterative deconvolution in the time domain, and then an initial model consisting of a series of 2-km-thick isotropic layers to fit velocity models, and thus determine an overall statistical solution by means of the bootstrap resampling technique. This statistical solution is then regarded as a new initial model to adjust high-frequency PRFs. Hence, the same resampling process is executed again to estimate the optimal S-wave velocity structure below each station. The results provide an accurate 3D image of the crust and uppermost mantle in the SE margin of Tibet. We infer a wide intra-crustal low-velocity zone that varies laterally and in depth, which is thinner or even absent in the most southern part of Yunnan. Our hypothesis is that this low-velocity zone is the result of the accumulation of lower crustal flow coming from central Tibet. Furthermore, we show that this lower crustal flow extends largely through the Sichuan–Yunnan diamond-shaped block, and that there are significant variations in both crustal velocity structure and deformation mechanism across the great strike-slip faults of the Jinshajiang–Red River and Xiaojiang fault systems.